1. Field of the Invention
The present invention relates to an image projection system operable within the visible spectrum which includes a polarizing beam splitter which reflects one linear polarization of light and transmits the other. More particularly, the present invention relates to such an image projection system with a compact, lightweight beam splitter that is comprised of a plurality of elongated, reflective elements for interacting with the electromagnetic waves of the source light to generally transmit or pass one polarization of light, and reflect the other polarization.
2. Prior Art
Polarized light is necessary in certain applications, such as projection liquid crystal displays (LCD). Such a display is typically comprised of a light source; optical elements, such as lenses to gather and focus the light; a polarizer that transmits one polarization of the light to the liquid crystal array; a liquid crystal array for manipulating the polarization of the light to encode image information thereon; means for addressing each pixel of the array to either change or retain the polarization; a second polarizer (called an analyzer) to reject the unwanted light from the selected pixels; and a screen upon which the image is focused.
It is possible to use a single polarizing beam splitter (PBS) to serve both as the first polarizer and the second polarizer (analyzer). If the liquid crystal array is reflective, for example a Liquid Crystal On Silicon (LCOS) light valve, it can reflect the beam that comes from the polarizer directly back to the polarizer after encoding the image by modifying the polarization of selected pixels. Such a system was envisioned by Takanashi (U.S. Pat. No. 5,239,322). The concept was elaborated by Fritz and Gold (U.S. Pat. No. 5,513,023). These similar approaches would provide important advantages in optical layout and performance. Neither, however, has been realized in practice because of deficiencies in conventional polarizing beam splitters. The disadvantages of using conventional polarizing beam splitters in projection liquid crystal displays includes images that are not bright, have poor contrast, and have non-uniform color balance or non-uniform intensity (due to non-uniform performance over the light cone). In addition, many conventional polarizing beam splitters are short-lived because of excessive heating, and are very expensive.
In order for such an image projection system to be commercially successful, it must deliver images which are significantly better than the images provided by conventional cathode ray tube (CRT) television displays because it is likely that such a system will be more expensive than conventional CRT technology. Therefore, the image projection system must provide (1) bright images with the appropriate colors or color balance; (2) have good image contrast; and (3) be as inexpensive as possible. An improved polarizing beam splitter (PBS) is an important part of achieving this goal because the PBS is a limiting component which determines the potential performance of the display system.
The PBS characteristics which significantly affect the display performance are (1) the angular aperture, or the f-number, at which the polarizer can function; (2) the absorption, or energy losses, associated with the use of the PBS; and (3) the durability of the PBS. In optics, the angular aperture or f-number describes the angle of the light cone which the PBS can use and maintain the desired performance level. Larger cones, or smaller f-numbers, are desired because the larger cones allow for more light to be gathered from the light source, which leads to greater energy efficiency and more compact systems.
The absorption and energy losses associated with the use of the PBS obviously affect the brightness of the system since the more light lost in the optics, the less light remains which can be projected to the view screen. In addition, the amount of light energy which is absorbed by the polarizer will affect its durability, especially in such image projection systems in which the light passing through the optical system is very intense, on the order of watts per square centimeter. Light this intense can easily damage common polarizers, such as Polaroid sheets. In fact, the issue of durability limits the polarizers which can be used in these applications.
Durability is also important because the smaller and lighter the projection system can be made, the less expensive and more desirable is the product. To accomplish this goal, however, the light intensity must be raised even higher, further stressing the PBS, and shortening its useful life.
A problematic disadvantage of conventional PBS devices is poor conversion efficiency, which is the primary critical performance factor in displays. Conversion efficiency is a measure describing how much of the electrical power required by the light source is translated into light intensity power on the screen or panel that is observed by people viewing it. It is expressed as the ratio of total light power on the screen divided by the electrical power required by the light source. The conventional units are lumens per watt. A high ratio is desirable for a number of reasons. For example, a low conversion efficiency will require a brighter light source, with its accompanying larger power supply, excess heat, larger enclosures and cabinet, etc. In addition, all of these consequences of low conversion efficiency raise the cost of the projection system.
A fundamental cause of low conversion efficiency is poor optical efficiency, which is directly related to the f-number of the optical system. A system which has an f-number which is half the f-number of an otherwise equivalent system has the potential to be four times as efficient in gathering light from the light source. Therefore, it is desirable to provide an improved polarizing beam splitter (PBS) which allows more efficient harvesting of light energy by offering a significantly smaller potential f-number (larger angular aperture), and therefore increases the conversion efficiency, as measured in lumens/watt.
There are several reasons for the poor performance of conventional polarizing beam splitters with respect to conversion efficiency when they are used as beam splitters in projection systems. First, current beam splitters work poorly if the light does not strike them at a certain angle (or at least, within a narrow cone of angles about this principal angle of incidence). Deviation of the principal ray from this angle causes each type of polarizing beam splitter to degrade the intensity, the purity of polarization, and/or the color balance. This applies to the beam coming from the light source as well as to the beam reflected from the liquid crystal array. This principal angle depends upon the design and construction of the PBS as well as the physics of the polarization mechanism employed in these various beam splitters. Currently available polarizing beam splitters are not capable of operating efficiently at angles far from their principal polarizing angles in the visible portion of the electromagnetic spectrum. This restriction makes it impossible to implement certain promising optical layouts and commercially promising display designs.
Even if the principal ray strikes the polarizer at the best angle for separating the two polarizations, the other rays cannot diverge far from this angle or their visual qualities will be degraded. This is a serious deficiency in a display apparatus because the light striking the polarizer must be strongly convergent or divergent to make efficient use of the light emitted by typical light sources. This is usually expressed as the f-number of the optical system. For a single lens, the f-number is the ratio of the aperture to the focal length. For optical elements in general, the F-number is defined as EQU F/#=1/(2 n sin .theta.)
where n is the refractive index of the space within which the optical element is located, and .theta. is the half cone angle. The smaller the F-number, the greater the radiant flux, .phi..sub.c, collected by the lens, and the more efficient the device will be for displaying a bright image. The radiant flux increases as the inverse square of the F/#. In an optical train, the optical element with the largest F/# will be the limiting factor in its optical efficiency. For displays using traditional polarizers, the limiting element is nearly always the polarizer, and thus the PBS limits the conversion efficiency. It would clearly be beneficial to develop a type of PBS that has a smaller F/# than any that are currently available.
Because traditional polarizers with small F/#s have not been available, designers typically have addressed the issue of conversion efficiency by specifying a smaller, brighter light source. Such sources, typically arc lamps, are available, but they require expensive power supplies that are heavy, bulky, and need constant cooling while in operation. Cooling fans cause unwanted noise and vibration. These features are detrimental to the utility of projectors and similar displays. Again, a PBS with a small F/# would enable efficient gathering of light from low-power, quiet, conventional light sources.
Another key disadvantage of conventional polarizing beam splitters is a low extinction, which results in poor contrast in the image. Extinction is the ratio of the light transmitted through the polarizer of the desired polarization to the light rejected of the undesired polarization. In an efficient display, this ratio must be maintained at a minimum value over the entire cone of light passing through the PBS. Therefore, it is desirable to provide a polarizing beam splitter which has a high extinction ratio resulting in a high contrast image.
A third disadvantage of conventional polarizing beam splitters is a non-uniform response over the visible spectrum, or poor color fidelity. The result is poor color balance which leads to further inefficiency in the projection display system as some light from the bright colors must be thrown away to accommodate the weaknesses in the polarizing beam splitter. Therefore, it is desirable to provide an improved polarizing beam splitter that has a uniform response over the visible spectrum, (or good color fidelity) giving an image with good color balance with better efficiency. The beam splitter must be achromatic rather than distort the projected color, and it must not allow crosstalk between the polarizations because this degrades image acuity and contrast. These characteristics must apply over all portions of the polarizer and over all angles of light incidence occurring at the polarizer. The term spathic has been coined (R. C. Jones, Jour. Optical Soc. Amer. 39, 1058, 1949) to describe a polarizer that conserves cross-sectional area, solid angle, and the relative intensity distribution of wavelengths in the polarized beam. A PBS that serves as both a polarizer and analyzer must be spathic for both transmission and reflection, even in light beams of large angular aperture.
A fourth disadvantage of conventional polarizing beam splitters is poor durability. Many conventional polarizing beam splitters are subject to deterioration caused by excessive heating and photochemical reactions. Therefore, it is desirable to provide an improved polarizing beam splitter that can withstand an intense photon flux for thousands of hours without showing signs of deterioration. In addition, it is desirable to provide a polarizing beam splitter that is amenable to economical large scale fabrication.
The need to meet these, and other, criteria has resulted in only a few types of polarizers finding actual use in a projection system. Many attempts have been made to incorporate both wide angular aperture and high fidelity polarization into the same beam splitting device. The relative success of these efforts is described below. Thin film interference filters are the type of polarizer cited most frequently in efforts to make a polarizing beam splitter that is also used as an analyzer. MacNeille was the first to describe such a polarizer that was effective over a wide spectral range (U.S. Pat. No. 2,403,731). It is composed of thin-film multi-layers set diagonally to the incident light, typically within a glass cube, so it is bulky and heavy compared to a sheet polarizer. What is more, it must be designed for a single angle of incidence, usually 45.degree., and its performance is poor if light is incident at angles different from this by even as little as 2.degree.. Others have improved on the design (e.g. J. Mouchart, J. Begel, and E. Duda, Applied Optics 28, 2847-2853, 1989; and L. Li and J. A. Dobrowolski, Applied Optics 13, 2221-2225, 1996). All of them found it necessary to seriously reduce the wavelength range if the angular aperture is to be increased. This can be done in certain designs (U.S. Pat. Nos. 5,658,060 and 5,798,819) in which the optical design divides the light into appropriate color bands before it arrives at the polarizing beam splitter. In this way, it is possible to reduce the spectral bandwidth demands on the beam splitter and expand its angular aperture, but the additional components and complexity add significant cost, bulk, and weight to the system.
Even so, these improved beam splitter cubes are appearing on the market, and are currently available from well known vendors such as Balzers and OCLI. They typically offer an F/# of f/2.5-f/2.8, which is a significant improvement over what was available 2 years ago, but is still far from the range of F/1.2-F/2.0 which is certainly within reach of the other key components in optical projection systems. Reaching these f-numbers has the potential to improve system efficiency by as much as a factor of 4. They would also enable the projection display engineer to make previously impossible design trade-offs to achieve other goals, such as reduced physical size and weight, lower cost, etc.
In a technology far from visible optics, namely radar, wire grids have been used successfully to polarize long wavelength radar waves. These wire grid polarizers have also been used as reflectors. They are also well known as optical components in the infrared (IR), where they are used principally as transmissive polarizer elements.
Although it has not been demonstrated, some have postulated possible use of a wire grid polarizer in display applications in the visible portion of the spectrum. For example, Grinberg (U.S. Pat. No. 4,688,897) suggested that a wire grid polarizer serve as both a reflector and an electrode (but not simultaneously as an analyzer) for a liquid crystal display.
Others have posed the possible use of a wire grid polarizer in place of a dichroic polarizer to improve the efficiency of virtual image displays (see U.S. Pat. No. 5,383,053). The need for contrast or extinction in the grid polarizer, however, is explicitly dismissed, and the grid is basically used as a polarization sensitive beam steering device. It does not serve the purpose of either an analyzer, or a polarizer, in the U.S. Pat. No. 5,383,053. It is also clear from the text that a broadband polarizing cube beam splitter would have served the purpose as well, if one had been available. This technology, however, is dismissed as being too restricted in acceptance angle to even be functional, as well as prohibitively expensive.
Another patent (U.S. Pat. No. 4,679,910) describes the use of a grid polarizer in an imaging system designed for the testing of IR cameras and other IR instruments. In this case, the application requires a beam splitter for the long wavelength infra-red, in which case a grid polarizer is the only practical solution. The patent does not suggest utility for the visible range or even mention the need for a large angular aperture. Neither does it address the need for efficient conversion of light into a viewable image, nor the need for broadband performance.
Other patents also exist for wire-grid polarizers in the infrared portion of the spectrum (U.S. Pat. Nos. 4,514,479, 4,743,093; and 5,177,635, for example). Except for the exceptions just cited, the emphasis is solely on the transmission performance of the polarizer in the IR spectrum.
These references demonstrate that it has been known for many years that wire-grid arrays can function generally as polarizers. Nevertheless, they apparently have not been proposed and developed for image projection systems. One possible reason that wire grid polarizers have not been applied in the visible spectrum is the difficulty of manufacture. U.S. Pat. No. 4,514,479 teaches a method of holographic exposure of photoresist and subsequent etching in an ion mill to make a wire grid polarizer for the near infrared region; in U.S. Pat. No. 5,122,907, small, elongated ellipsoids of metal are embedded in a transparent matrix that is subsequently stretched to align their long axes of the metal ellipsoids to some degree. Although the transmitted beam is polarized, the device does not reflect well. Furthermore, the ellipsoid particles have not been made small enough to be useful in the visible part of the electromagnetic spectrum. Accordingly, practical applications have been generally limited to the longer wavelengths of the IR spectrum.
Another prior art polarizer achieves much finer lines by grazing angle evaporative deposition (U.S. Pat. No. 4,456,515). Unfortunately, the lines have such small cross sections that the interaction with the visible light is weak, and so the optical efficiency is too poor for use in the production of images. As in several of these prior art efforts, this device has wires with shapes and spacings that are largely random. Such randomness degrades performance because regions of closely spaced elements do not transmit well, and regions of widely spaced elements have poor reflectance. The resulting degree of polarization (extinction) is less than maximal if either or both of these effects occur, as they surely must if placement has some randomness to it.
For perfect (and near perfect) regularity, the mathematics developed for gratings apply well. Conversely, for random wires (even if they all have the same orientation) the theory of scattering provides the best description. Scattering from a single cylindrical wire has been described (H. C. Van de Hulst, Light Scattering by Small Particles, Dover, 1981). The current random-wire grids have wires embedded throughout the substrate. Not only are the positions of the wires somewhat random, but the diameters are as well. It is clear that the phases of the scattered rays will be random, so the reflection will not be strictly specular and the transmission will not retain high spacial or image fidelity. Such degradation of the light beam would prevent its use for transfer of well resolved, high-information density images.
Nothing in the prior art indicates or suggests that an ordered array of wires can or should be made to operate over the entire visible range as a spathic PBS, at least at the angles required when it serves both as a polarizer and analyzer. Indeed, the difficulty of making the narrow, tall, evenly spaced wires that are required for such operation has been generously noted (see Zeitner, et. al. Applied Optics, 38, 11 pp. 2177-2181 (1999), and Schnabel, et. al., Optical Engineering 38,2 pp. 220-226 (1999)). Therefore, it is not surprising that the prior art for image projection similarly makes no suggestion for use of a spathic PBS as part of a display device.
Tamada and Matsumoto (U.S. Pat. No. 5,748,368) disclose a wire grid polarizer that operates in both the infrared and a portion of the visible spectrum; however, it is based on the concept that large, widely spaced wires will create resonance and polarization at an unexpectedly short wavelength in the visible. Unfortunately, this device works well only over a narrow band of visible wavelengths, and not over the entire visible spectrum. It is therefore not suitable for use in producing images in full color. Accordingly, such a device is not practical for image display because a polarizer must be substantially achromatic for an image projection system.
Another reason wire grid polarizers have been overlooked is the common and long standing belief that the performance of a typical wire grid polarizer becomes degraded as the light beam's angle of incidence becomes large (G. R. Bird and M. Parrish, Jr., "The Wire Grid as a Near-Infrared Polarizer," J. Opt. Soc. Am., 50, pp. 886-891, (1960); the Handbook of Optics, Michael Bass, Volume II, p. 3-34, McGraw-Hill (1995)). There are no reports of designs that operate well for angles beyond 35.degree. incidence in the visible portion of the spectrum. Nor has anyone identified the important design factors that cause this limitation of incidence angle. This perceived design limitation becomes even greater when one realizes that a successful beam splitter will require suitable performance in both transmission and reflection simultaneously.
This important point deserves emphasis. The extant literature and patent history for wire grid polarizers in the IR and the visible spectra has almost entirely focused on their use as transmission polarizers, and not on reflective properties. Wire grid polarizers have been attempted and reported in the technical literature for decades, and have become increasingly common since the 1960s. Despite the extensive work done in this field, there is very little, if any, detailed discussion of the production and use of wire grid polarizers as reflective polarizers, and nothing in the literature concerning their use as both transmissive and reflective polarizers simultaneously, as would be necessary in a spathic polarizing beam splitter for use in imaging devices. From the lack of discussion in the literature, a reasonable investigator would conclude that any potential use of wire grid polarizers as broadband visible beam splitters is riot apparent, or that it was commonly understood by the technical community that their use in such an application was not practical.
Because the conventional polarizers described above were the only ones available, it was impossible for Takanashi (U.S. Pat. No. 5,239,322) to reduce his projection device to practice with anything but the most meager results. No polarizer was available which supplied the performance required for the Takanashi invention, namely, achromaticity over the visible part of the spectrum, wide angular acceptance, low losses in transmission and reflection of the desired light polarizations, and good extinction ratio.
There are several important features of an image display system which require specialized performance of transmission and reflection properties. For a projector, the product of p-polarization transmission and s-polarization reflection (R.sub.s T.sub.p) must be large if source light is to be efficiently placed on the screen. On the other hand, for the resolution and contrast needed to achieve high information density on the screen, it is important that the converse product (R.sub.p T.sub.s) be very small (i.e. the transmission of s-polarized light multiplied by the reflection of p-polarized light must be small).
Another important feature is a wide acceptance angle. The acceptance angle must be large if light gathering from the source, and hence the conversion efficiency, is maximized. It is desirable that cones of light (either diverging or converging) with half-angles greater than 20.degree. be accepted.
An important consequence of the ability to accept larger light cones and work well at large angles is that the optical design of the imaging system is no longer restricted. Conventional light sources can be then be used, bringing their advantages of low cost, cool operation, small size, and low weight. A wide range of angles makes it possible for the designer to position the other optical elements in favorable positions to improve the size and operation of the display.
Another important feature is size and weight. The conventional technology requires the use of a glass cube. This cube imposes certain requirements and penalties on the system. The requirements imposed include the need to deal with thermal loading of this large piece of glass and the need for high quality materials without stress birefringence, etc., which impose additional cost. In addition, the extra weight and bulk of the cube itself poses difficulties. Thus, it is desirable that the beam splitter not occupy much volume and does not weigh very much.
Another important feature is robustness. Modern light sources generate very high thermal gradients in the polarizer immediately after the light is switched on. At best, this can induce thermal birefringence which causes cross talk between polarizations. What is more, the long duration of exposure to intense light causes some materials to change properties (typically yellowing from photo-oxidation). Thus, it is desirable for the beam splitter to withstand high temperatures as well as long periods of intense radiation from light sources.
Still another important feature is uniform extinction (or contrast) performance of the beam splitter over the incident cone of light. A McNeille-type thin film stack polarizer produces polarized light due to the difference in reflectivity of S-polarized light as opposed to P-polarized light. Since the definition of S and P polarization depends on the plane of incidence of the light ray, which changes orientation within the cone of light incident on the polarizer, a McNeille-type polarizer does not work equally well over the entire cone. This weakness in McNeille-type polarizers is well known. It must be addressed in projection system design by restricting the angular size of the cone of light, and by compensation elsewhere in the optical system through the use of additional optical components. This fundamental weakness of McNeille prisms raises the costs and complexities of current projection systems, and limits system performance through restrictions on the f-number or optical efficiency of the beam splitter.
Other important features include ease of alignment. Production costs and maintenance are both directly affected by assembly criteria. These costs can be significantly reduced with components which do not require low tolerance alignments.
Therefore, it would be advantageous to develop an image projection system capable of providing bright images and good image contrast, and which is inexpensive. It would also be advantageous to develop an image projection system with a polarizing beam splitter capable of utilizing divergent light (or having a smaller F/#), capable of efficient use of light energy or with high conversion efficiency, and which is durable. It would also be advantageous to develop an image projection system with a polarizing beam splitter having a high extinction ratio, uniform response over the visible spectrum, good color fidelity, that is spathic, robust and capable of resisting thermal gradients. It would also be advantageous to develop an image projection system with a polarizing beam splitter capable of being positioned at substantially any incidence angle so that significant design constraints are not imposed on the image projection system, but substantial design flexibility is permitted. It would also be advantageous to develop an image projection system with a polarizing beam splitter which efficiently transmits p-polarized light and reflects s-polarized light across all angles in the entire cone of incident light. It would also be advantageous to develop an image projection system with a polarizing beam splitter which is light-weight and compact. It would also be advantageous to develop an image projection system with a polarizing beam splitter which is easy to align. Combining all of these features in a single projection device would offer a significant advance within the state of the art.